Reproduce “SearchDiffusion: Information-Guided Diffusion Planning for Object Search” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Sausar Karaf, Dmitrii Plotnikov, Alexey Kotcov, Dzmitry Tsetserukou

SearchDiffusion plans object searches by predicting a dense utility map encoding accessibility, useful visibility, and semantic search progress. It uses the map to condition trajectory generation and guide refinement; validity checks and analytical scoring select executable plans. In simulations, it detected targets in 95.6% of tasks versus 85.6% for fine-tuned MapEx, but median planning time was higher: 307.3 versus 44.6 milliseconds.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.